Final Project DSA
Nolan Jolicoeur
12/6/2021
- First I am importing the necessary packages.
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(ggplot2)
library(sqldf)## Loading required package: gsubfn## Loading required package: proto## Loading required package: RSQLitelibrary(RSQLite)
library(devtools)## Loading required package: usethislibrary(ggpubr)
library(tidyverse)## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.1 ──## ✓ tibble 3.1.5 ✓ purrr 0.3.4
## ✓ tidyr 1.1.4 ✓ stringr 1.4.0
## ✓ readr 2.0.2 ✓ forcats 0.5.1## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## x dplyr::filter() masks stats::filter()
## x dplyr::lag() masks stats::lag()library(knitr)Summary of how fantasy football works People draft teams from a pool of real NFL players Points are gained on a weekly basis depending on how their players perform Point System:
Passing:
- Touchdown = 4
points Yard = 0.04 points
Running:
Touchdown = 6 points
Yard = 0.1 points
Receiving:
Touchdown = 6 points
Yard = 0.1 points
Catch = 1 points
Fantasy team consist of quarterbacks, running backs, and wide receivers. In this project I tested the strength of the relationships between different statistics and fantasy points We want to find which fantasy football stats matter the most and which calculated stats might also relate to fantasy points We also can sum up the fantasy points for all the NFL teams to find out if fantasy football success relates to actual team success There will be 2 parts to this project. Part 1 is team based stats and part 2 is a more specific dive into stats by position group.
#Data is through November 29, 2021
df <- read.csv("~/Downloads/nfl_pass_rush_receive_raw_data (1).csv")
#This data frame contains data from the start of 2019 to the current day in the NFL- First I have to create a data frame of team records in order to do the first part.
- I could not take one from the internet because it had to much the other data I imported and the ones on the internet update weekly.
team = c("ARI", "ATL", "BAL", "BUF", "CAR", "CHI", "CIN", "CLE", "DAL", "DEN", "DET", "GNB", "HOU", "IND", "JAX", "KAN", "LVR", "LAC", "LAR", "MIA", "MIN", "NWE", "NOR", "NYG", "NYJ", "PHI", "PIT", "SFO", "SEA", "TAM", "TEN", "WAS")
wins = c(9,5,8,7,5,4,7,6,7,6,0,9,2,6,2,7,6,6,7,5,5,8,5,4,3,5,5,6,3,8,8,5)
losses = c(2,6,3,4,7,7,4,6,4,5,10,3,9,6,9,4,5,5,4,7,6,4,6,7,8,7,5,5,8,3,4,6)
ties = c(0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0)
points_for = c(310,199,263,326,236,179,309,254,326,228,174,283,164,340,173,281,259,273,299,234,281,336,257,202,199,304,224,280,209,347,304,229)
points_against = c(202,302,240,182,253,254,226,267,250,196,289,242,292,283,283,250,295,293,263,279,276,190,249,253,334,273,267,248,226,253,290,282)
team_records = data.frame(team, wins, losses, ties, points_for, points_against)Next I use SQL to organize the data and combine data sets using the ‘team’ column for a left join.
This will allow me to compare team records and points to fantasy points.
Fantasy_PTS_By_Team = sqldf("select team, sum(Total_DKP)
from df
group by team")
Fantasy_PTS_By_Team = rename(Fantasy_PTS_By_Team, 'Total_DKP' = 'sum(Total_DKP)')
Combined_Team_Stats = sqldf("select team_records.team, team_records.wins, team_records.losses, team_records.ties, team_records.points_for, team_records.points_against, Fantasy_PTS_By_Team.Total_DKP
from team_records
left join Fantasy_PTS_By_Team on team_records.team = Fantasy_PTS_By_Team.team")
Fantasy_PTS_Against_Team = sqldf("select Opponent_abbrev, sum(Total_DKP)
from df
group by Opponent_abbrev")
Fantasy_PTS_Against_Team = rename(Fantasy_PTS_Against_Team, 'Total_DKP' = 'sum(Total_DKP)')
Fantasy_PTS_Against_Team = rename(Fantasy_PTS_Against_Team, 'team' = 'Opponent_abbrev')
Combined_Points_Against = sqldf("select team_records.team, team_records.wins, team_records.losses, team_records.ties, team_records.points_for, team_records.points_against, Fantasy_PTS_Against_Team.Total_DKP
from team_records
left join Fantasy_PTS_Against_Team on team_records.team = Fantasy_PTS_Against_Team.team")- This is a set of graphs, correlations, and summaries for the team statistics that I am comparing.
Points_Per_Win <- ggplot(Combined_Team_Stats, aes(x = Total_DKP, y = wins))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Fantasy Points and Wins", x = "Fantasy Points", y = "Wins")+
geom_point()
Points_Per_Win## `geom_smooth()` using formula 'y ~ x'
Fantasy_PTS_Per_Actual_Point <- ggplot(Combined_Team_Stats, aes(x = Total_DKP, y = points_for))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Fantasy Points and Points For", x = "Fantasy Points", y = "Points For")+
geom_point()
Fantasy_PTS_Per_Actual_Point## `geom_smooth()` using formula 'y ~ x'
Fantasy_PTS_Per_Point_Against <- ggplot(Combined_Points_Against, aes(x = Total_DKP, y = points_against))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Fantasy Points Against and Points For", x = "Fantasy Points Against", y = "Points For")+
geom_point()
Fantasy_PTS_Per_Point_Against## `geom_smooth()` using formula 'y ~ x'
cor_wins = cor.test(Combined_Team_Stats$Total_DKP, Combined_Team_Stats$wins, method = "pearson")
cor_wins##
## Pearson's product-moment correlation
##
## data: Combined_Team_Stats$Total_DKP and Combined_Team_Stats$wins
## t = 6.6849, df = 30, p-value = 2.092e-07
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.5817269 0.8838251
## sample estimates:
## cor
## 0.7735171model_wins = lm(points_for~Total_DKP, Combined_Team_Stats)
summary(model_wins)##
## Call:
## lm(formula = points_for ~ Total_DKP, data = Combined_Team_Stats)
##
## Residuals:
## Min 1Q Median 3Q Max
## -38.366 -17.102 -5.118 16.970 60.128
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -99.20128 35.03917 -2.831 0.0082 **
## Total_DKP 0.34691 0.03367 10.304 2.27e-11 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 25.63 on 30 degrees of freedom
## Multiple R-squared: 0.7797, Adjusted R-squared: 0.7723
## F-statistic: 106.2 on 1 and 30 DF, p-value: 2.267e-11# y = 0.0118x - 6.6243 R-Squared = 0.5849 (Good)
cor_points_for = cor.test(Combined_Team_Stats$Total_DKP, Combined_Team_Stats$points_for, method = "pearson")
cor_points_for##
## Pearson's product-moment correlation
##
## data: Combined_Team_Stats$Total_DKP and Combined_Team_Stats$points_for
## t = 10.304, df = 30, p-value = 2.267e-11
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.7720058 0.9417371
## sample estimates:
## cor
## 0.8830013model_points_for = lm(wins~Total_DKP, Combined_Team_Stats)
summary(model_points_for)##
## Call:
## lm(formula = wins ~ Total_DKP, data = Combined_Team_Stats)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.5947 -0.9079 0.0239 0.7526 2.2182
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -6.624356 1.843185 -3.594 0.00115 **
## Total_DKP 0.011839 0.001771 6.685 2.09e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.348 on 30 degrees of freedom
## Multiple R-squared: 0.5983, Adjusted R-squared: 0.5849
## F-statistic: 44.69 on 1 and 30 DF, p-value: 2.092e-07# y = 0.3469x - 99.2013 R-Squared = 0.7723
cor_points_against = cor.test(Combined_Points_Against$Total_DKP, Combined_Points_Against$points_against, method = "pearson")
cor_points_against##
## Pearson's product-moment correlation
##
## data: Combined_Points_Against$Total_DKP and Combined_Points_Against$points_against
## t = 6.3442, df = 30, p-value = 5.349e-07
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.5546285 0.8747487
## sample estimates:
## cor
## 0.7569346model_points_against = lm(points_against~Total_DKP, Combined_Points_Against)
summary(model_points_against)##
## Call:
## lm(formula = points_against ~ Total_DKP, data = Combined_Points_Against)
##
## Residuals:
## Min 1Q Median 3Q Max
## -55.896 -14.128 1.545 14.001 47.603
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -35.32768 46.53972 -0.759 0.454
## Total_DKP 0.28502 0.04493 6.344 5.35e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 22.9 on 30 degrees of freedom
## Multiple R-squared: 0.573, Adjusted R-squared: 0.5587
## F-statistic: 40.25 on 1 and 30 DF, p-value: 5.349e-07# y = 0.2850x - 35.3277 R-Squared = 0.5587
team_table = c(cor_wins$estimate, cor_points_for$estimate, cor_points_against$estimate)
row_names3 = c("Wins", "Points For", "Points Against")
team_table = as.data.frame(team_table, row_names3)
team_table## team_table
## Wins 0.7735171
## Points For 0.8830013
## Points Against 0.7569346Now I am going to compare how the opponents defense affects fantasy points by position
I will do this by extracting the 5 best and 5 worst fantasy defenses and seeing how many points were scored against them on average by position group.
worst_teams = team_records[order(points_against),]
bottom_ten = tail(worst_teams, 5)
bottom_ten_teams = bottom_ten$team
bottom_ten_teams_df = as.data.frame(bottom_ten_teams)
Vs_bottom_ten = sqldf("select df.pos, df.pass_att, df.rush_att, df.rec, df.team, df.Opponent_abbrev, df.Total_DKP, bottom_ten_teams_df.bottom_ten_teams
from df
left join bottom_ten_teams_df on df.Opponent_abbrev = bottom_ten_teams_df.bottom_ten_teams")
Vs_bottom_ten = Vs_bottom_ten[complete.cases(Vs_bottom_ten),]
QB_vs_bottom_ten = sqldf("select pos, avg(Total_DKP)
from Vs_bottom_ten
where pass_att >= 10 and pos = 'QB'")
RB_vs_bottom_ten = sqldf("select pos, avg(Total_DKP)
from Vs_bottom_ten
where rush_att >= 5 and pos = 'RB'")
WR_TE_vs_bottom_ten = sqldf("select pos, avg(Total_DKP)
from Vs_bottom_ten
where rec >= 3 and pos ='WR' or pos ='TE'")
top_ten = head(worst_teams, 5)
top_ten_teams = top_ten$team
top_ten_teams_df = as.data.frame(top_ten_teams)
Vs_top_ten = sqldf("select df.pos, df.team, df.Opponent_abbrev, df.pass_att, df.rush_att, df.rec, df.Total_DKP, top_ten_teams_df.top_ten_teams
from df
left join top_ten_teams_df on df.Opponent_abbrev = top_ten_teams_df.top_ten_teams")
Vs_top_ten = Vs_top_ten[complete.cases(Vs_top_ten),]
QB_vs_top_ten = sqldf("select pos, avg(Total_DKP)
from Vs_top_ten
where pass_att >= 10 and pos = 'QB'")
RB_vs_top_ten = sqldf("select pos, avg(Total_DKP)
from Vs_top_ten
where rush_att >= 5 and pos = 'RB'")
WR_TE_vs_top_ten = sqldf("select pos, avg(Total_DKP)
from Vs_top_ten
where rec >= 3 and pos ='WR' or pos ='TE'")
vs_top_teams = c(QB_vs_top_ten$`avg(Total_DKP)`, RB_vs_top_ten$`avg(Total_DKP)`, WR_TE_vs_top_ten$`avg(Total_DKP)`)
vs_worst_teams = c(QB_vs_bottom_ten$`avg(Total_DKP)`, RB_vs_bottom_ten$`avg(Total_DKP)`, WR_TE_vs_bottom_ten$`avg(Total_DKP)`)
row_names4 = c("QB Fantasy Points", "RB Fantasy Points", "WR/TE Fantasy Points")
positional_top_table = as.data.frame(vs_top_teams,row_names4)
positional_worst_table = as.data.frame(vs_worst_teams, row_names4)
knitr::kable(list(positional_top_table,positional_worst_table))
|
|
Part 1 conclusion:
Fantasy points does relate to wins but has a stronger relationship to points scored than wins.
On average, players do put up more fantasy points when playing against bottom 5 teams vs top 5 teams as far as points allowed.
There is a correlation between team success and fantasy points.
Next I will organize the data by position starting with QB.
QB_df1 <- sqldf("select pos, player, team, pass_cmp, pass_att, pass_yds, pass_td, pass_int, pass_sacked, pass_sacked_yds, pass_long, pass_rating, comb_pass_rush_play, rush_scrambles, comb_pass_play, comb_rush_play, pass_poor_throws, pass_blitzed, pass_hurried, Total_DKP
from df
where pos = 'QB' and pass_att > 15")
QB_totals = sqldf("select pos, player, team, avg(pass_cmp), avg(pass_att), avg(pass_yds), avg(pass_td), avg(pass_int), avg(pass_sacked), avg(pass_long), avg(pass_rating), avg(rush_scrambles), avg(comb_pass_play), avg(comb_rush_play), avg(pass_poor_throws), avg(pass_blitzed), avg(pass_hurried), avg(Total_DKP)
from QB_df1
group by player")pass_att must be greater than 15 because we are trying to exclude QB’s who were subbed into the end of a game that was a blowout.
They will be automatic outlines and will skew the data.
Now we must rename the columns so we can use them in SQL.
QB_totals = rename(QB_totals, 'AVG_Cmp' = 'avg(pass_cmp)')
QB_totals = rename(QB_totals, 'AVG_Pass_Att' = 'avg(pass_att)')
QB_totals = rename(QB_totals, 'AVG_Pass_Yards' = 'avg(pass_yds)')
QB_totals = rename(QB_totals, 'AVG_Pass_TD' = 'avg(pass_td)')
QB_totals = rename(QB_totals, 'AVG_Pass_Sacked' = 'avg(pass_sacked)')
QB_totals = rename(QB_totals, 'AVG_Pass_Long' = 'avg(pass_long)')
QB_totals = rename(QB_totals, 'AVG_Pass_Rating' = 'avg(pass_rating)')
QB_totals = rename(QB_totals, 'AVG_INT' = 'avg(pass_int)')
QB_totals = rename(QB_totals, 'AVG_Pass_Play' = 'avg(comb_pass_play)')
QB_totals = rename(QB_totals, 'AVG_Rush_Play' = 'avg(comb_rush_play)')
QB_totals = rename(QB_totals, 'AVG_Poor_Passes' = 'avg(pass_poor_throws)')
QB_totals = rename(QB_totals, 'AVG_Blitzed' = 'avg(pass_blitzed)')
QB_totals = rename(QB_totals, 'AVG_Hurried' = 'avg(pass_hurried)')
QB_totals = rename(QB_totals, 'AVG_Total_DKP' = 'avg(Total_DKP)')- Now I am going to graph each QB statistic in relation to fantasy points
pass_cmp <- ggplot(QB_totals, aes(x = AVG_Cmp, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Pass Completions and Average Fantasy Points", x = "Average Pass Completions", y = "Average Fantasy Points")+
geom_point()
pass_cmp## `geom_smooth()` using formula 'y ~ x'
pass_att <- ggplot(QB_totals, aes(x = AVG_Pass_Att, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Pass Attempts and Fantasy Points", x = "Average Pass Attempts", y = "Average Fantasy Points")+
geom_point()
pass_att## `geom_smooth()` using formula 'y ~ x'
pass_yds <- ggplot(QB_totals, aes(x = AVG_Pass_Yards, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Pass Yards and Fantasy Points", x = "Average Pass Yards", y = "Average Fantasy Points")+
geom_point()
pass_yds## `geom_smooth()` using formula 'y ~ x'
pass_td <- ggplot(QB_totals, aes(x = AVG_Pass_TD, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Pass Attempts and Fantasy Points", x = "Average Pass Attempts", y = "Average Fantasy Points")+
geom_point()
pass_td## `geom_smooth()` using formula 'y ~ x'
pass_sacked <- ggplot(QB_totals, aes(x = AVG_Pass_Sacked, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Times Sacked and Fantasy Points", x = "Average Times Sacked", y = "Average Fantasy Points")+
geom_point()
pass_sacked## `geom_smooth()` using formula 'y ~ x'
pass_long <- ggplot(QB_totals, aes(x = AVG_Pass_Long, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Longest Pass and Fantasy Points", x = "Average Longest Pass", y = "Average Fantasy Points")+
geom_point()
pass_long## `geom_smooth()` using formula 'y ~ x'
pass_rating <- ggplot(QB_totals, aes(x = AVG_Pass_Rating, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Passer Rating and Fantasy Points", x = "Average Passer Rating", y = "Average Fantasy Points")+
geom_point()
pass_rating## `geom_smooth()` using formula 'y ~ x'
comb_pass_play <- ggplot(QB_totals, aes(x = AVG_Pass_Play, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Number of Pass Plays and Fantasy Points", x = "Average Number of Pass Plays", y = "Average Fantasy Points")+
geom_point()
comb_pass_play## `geom_smooth()` using formula 'y ~ x'
comb_rush_play <- ggplot(QB_totals, aes(x = AVG_Rush_Play, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Number of Run Plays and Fantasy Points", x = "Average Number of Run Plays", y = "Average Fantasy Points")+
geom_point()
comb_rush_play## `geom_smooth()` using formula 'y ~ x'
poor_passes <- ggplot(QB_totals, aes(x = AVG_Poor_Passes, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Poor Passes and Fantasy Points", x = "Average Poor Passes", y = "Average Fantasy Points")+
geom_point()
poor_passes## `geom_smooth()` using formula 'y ~ x'
pass_blitzed <- ggplot(QB_totals, aes(x = AVG_Blitzed, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Passes Blitzed and Fantasy Points", x = "Average Passes Blitzed", y = "Average Fantasy Points")+
geom_point()
pass_blitzed## `geom_smooth()` using formula 'y ~ x'
pass_hurried <- ggplot(QB_totals, aes(x = AVG_Hurried, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Passes Hurried and Fantasy Points", x = "Average Passes Hurried", y = "Average Fantasy Points")+
geom_point()
pass_hurried## `geom_smooth()` using formula 'y ~ x'
- Next I will find the correlation between each statistic and fantasy points and create a table to display it.
cor_avg_cmp = cor.test(QB_totals$AVG_Cmp, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_att = cor.test(QB_totals$AVG_Pass_Att, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_yds = cor.test(QB_totals$AVG_Pass_Yards, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_td = cor.test(QB_totals$AVG_Pass_TD, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_sacked = cor.test(QB_totals$AVG_Pass_Sacked, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_long = cor.test(QB_totals$AVG_Pass_Long, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_rating = cor.test(QB_totals$AVG_Pass_Rating, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_int = cor.test(QB_totals$AVG_INT, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_play = cor.test(QB_totals$AVG_Pass_Play, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rush_play = cor.test(QB_totals$AVG_Rush_Play, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_poor_pass = cor.test(QB_totals$AVG_Poor_Passes, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_blitzed = cor.test(QB_totals$AVG_Blitzed, QB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_hurried = cor.test(QB_totals$AVG_Hurried, QB_totals$AVG_Total_DKP, method = "pearson")
QB_table = c(cor_avg_pass_att$estimate, cor_avg_cmp$estimate, cor_avg_pass_yds$estimate, cor_avg_pass_td$estimate, cor_avg_pass_sacked$estimate, cor_avg_pass_long$estimate, cor_avg_pass_rating$estimate, cor_avg_pass_int$estimate, cor_avg_pass_play$estimate, cor_avg_rush_play$estimate, cor_avg_poor_pass$estimate, cor_avg_blitzed$estimate, cor_avg_pass_hurried$estimate)
row_names2 = c("Average Attempts", "Average Completions", "Average Passing Yards", "Average Passing Touchdowns", "Average Times Sacked", "Average Longest Pass", "Average Passer Rating", "Average Interceptions", "Average Passing Plays", "Average Rushing Plays", "Average Poor Passes", "Average Times Blitzed", "Average Passes Hurried")
QB_table = as.data.frame(QB_table, row_names2)
QB_table## QB_table
## Average Attempts 0.577609660
## Average Completions 0.551552568
## Average Passing Yards 0.825130809
## Average Passing Touchdowns 0.887081580
## Average Times Sacked 0.103122367
## Average Longest Pass 0.487832083
## Average Passer Rating 0.713660641
## Average Interceptions -0.149726553
## Average Passing Plays 0.626841441
## Average Rushing Plays 0.302522184
## Average Poor Passes 0.295717554
## Average Times Blitzed -0.009947472
## Average Passes Hurried 0.053889982- I will now model the stats with the highest correlations to fantasy points.
#Passing touchdowns seems to have a higher correlation than passing yards to fantasy points.
model_pass_yds = lm(AVG_Total_DKP~AVG_Pass_Yards, QB_totals)
summary(model_pass_yds)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Pass_Yards, data = QB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -6.5063 -2.3218 -0.1601 1.1393 8.2428
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -3.290576 1.968680 -1.671 0.101
## AVG_Pass_Yards 0.086959 0.008337 10.430 3.01e-14 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.283 on 51 degrees of freedom
## Multiple R-squared: 0.6808, Adjusted R-squared: 0.6746
## F-statistic: 108.8 on 1 and 51 DF, p-value: 3.006e-14# y = 0.0869x + -3.291 R-Squared = 0.6746 (Above Average)
model_pass_td = lm(AVG_Total_DKP~AVG_Pass_TD, QB_totals)
summary(model_pass_td)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Pass_TD, data = QB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -5.886 -1.685 -0.411 1.740 7.455
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 6.9655 0.7992 8.716 1.13e-11 ***
## AVG_Pass_TD 7.2347 0.5272 13.724 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.683 on 51 degrees of freedom
## Multiple R-squared: 0.7869, Adjusted R-squared: 0.7827
## F-statistic: 188.3 on 1 and 51 DF, p-value: < 2.2e-16# y = 7.2347x + 6.9655 R-Squared = 0.7827 (Good)
#As far as stats not calculated into fantasy points, passer rating is the best predictor.
model_pass_rating = lm(AVG_Total_DKP~AVG_Pass_Rating, QB_totals)
summary(model_pass_rating)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Pass_Rating, data = QB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -11.2360 -2.5392 0.2442 1.4613 9.4898
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -6.67816 3.26116 -2.048 0.0457 *
## AVG_Pass_Rating 0.26759 0.03678 7.276 2e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.071 on 51 degrees of freedom
## Multiple R-squared: 0.5093, Adjusted R-squared: 0.4997
## F-statistic: 52.94 on 1 and 51 DF, p-value: 1.998e-09# y = 0.2676x -6.6782 R-Squared = 0.4997 (Average)- Next I will move onto RB’s and create a data frame for their statistics.
RB_df1 = sqldf("select pos, player, team, rush_att, rush_yds, rush_td, rush_long, targets, rec, rec_yds, rec_td, rec_long, fumbles_lost, designed_rush_att, rush_yds_before_contact, rush_yac, rush_broken_tackles, Total_DKP
from df
where pos = 'RB'")
RB_totals = sqldf("select pos, player, team, avg(rush_att), avg(rush_yds), avg(rush_td), avg(rush_long), avg(targets), avg(rec), avg(rec_yds), avg(rec_td), avg(rec_long), avg(fumbles_lost), avg(designed_rush_att), avg(rush_yds_before_contact), avg(rush_yac), avg(rush_broken_tackles), avg(Total_DKP)
from RB_df1
where rush_att > 5
group by player")- Now I must rename all the columns so I can use them in SQL.
RB_totals = rename(RB_totals, 'AVG_Total_DKP' = 'avg(Total_DKP)')
RB_totals = rename(RB_totals, 'AVG_Rush_Att' = 'avg(rush_att)')
RB_totals = rename(RB_totals, 'AVG_Rush_Yds' = 'avg(rush_yds)')
RB_totals = rename(RB_totals, 'AVG_Rush_TD' = 'avg(rush_td)')
RB_totals = rename(RB_totals, 'AVG_Rush_Long' = 'avg(rush_long)')
RB_totals = rename(RB_totals, 'AVG_Targets' = 'avg(targets)')
RB_totals = rename(RB_totals, 'AVG_Rec' = 'avg(rec)')
RB_totals = rename(RB_totals, 'AVG_Receiving_Yards' = 'avg(rec_yds)')
RB_totals = rename(RB_totals, 'AVG_Yds_Before_Contact' = 'avg(rush_yds_before_contact)')
RB_totals = rename(RB_totals, 'AVG_Yds_After_Contact' = 'avg(rush_yac)')
RB_totals = rename(RB_totals, 'AVG_Receiving_TD' = 'avg(rec_td)')
RB_totals = rename(RB_totals, 'AVG_Broken_Tackles' = 'avg(rush_broken_tackles)')- Now I will graph each stat in relation to fantasy points.
RB_rush_att <- ggplot(RB_totals, aes(x = AVG_Rush_Att, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Rushing Attempts and Fantasy Points", x = "Average Rushing Attempts", y = "Average Fantasy Points")+
geom_point()
RB_rush_att## `geom_smooth()` using formula 'y ~ x'
RB_rush_yds <- ggplot(RB_totals, aes(x = AVG_Rush_Yds, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Rushing Yards and Fantasy Points", x = "Average Rushing Yards", y = "Average Fantasy Points")+
geom_point()
RB_rush_yds## `geom_smooth()` using formula 'y ~ x'
RB_rush_td <- ggplot(RB_totals, aes(x = AVG_Rush_TD, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Rushing Touchdowns and Fantasy Points", x = "Average Rushing Touchdowns", y = "Average Fantasy Points")+
geom_point()
RB_rush_td## `geom_smooth()` using formula 'y ~ x'
RB_rush_long <- ggplot(RB_totals, aes(x = AVG_Rush_Long, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Longest Run and Fantasy Points", x = "Average Longest Run", y = "Average Fantasy Points")+
geom_point()
RB_rush_long## `geom_smooth()` using formula 'y ~ x'
RB_targets <- ggplot(RB_totals, aes(x = AVG_Targets, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Targets and Fantasy Points", x = "Average Targets", y = "Average Fantasy Points")+
geom_point()
RB_targets## `geom_smooth()` using formula 'y ~ x'
RB_rec <- ggplot(RB_totals, aes(x = AVG_Rec, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Receptions and Fantasy Points", x = "Average Receptions", y = "Average Fantasy Points")+
geom_point()
RB_rec## `geom_smooth()` using formula 'y ~ x'
RB_recieving_yds <- ggplot(RB_totals, aes(x = AVG_Receiving_Yards, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Recieving Yards and Fantasy Points", x = "Average Recieving Yards", y = "Average Fantasy Points")+
geom_point()
RB_recieving_yds## `geom_smooth()` using formula 'y ~ x'
RB_recieving_td <- ggplot(RB_totals, aes(x = AVG_Receiving_TD, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Recieving Touchdowns and Fantasy Points", x = "Average Recieving Touchdowns", y = "Average Fantasy Points")+
geom_point()
RB_recieving_td## `geom_smooth()` using formula 'y ~ x'
RB_yds_before_contact <- ggplot(RB_totals, aes(x = AVG_Yds_Before_Contact, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Yards Before Contact and Fantasy Points", x = "Average Yards Before Contact", y = "Average Fantasy Points")+
geom_point()
RB_yds_before_contact## `geom_smooth()` using formula 'y ~ x'
RB_yac <- ggplot(RB_totals, aes(x = AVG_Yds_After_Contact, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Yards After Contact and Fantasy Points", x = "Average Yards After Contact", y = "Average Fantasy Points")+
geom_point()
RB_yac## `geom_smooth()` using formula 'y ~ x'
RB_broken_tackles <- ggplot(RB_totals, aes(x = AVG_Broken_Tackles, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Broken Tackles and Fantasy Points", x = "Average Broken Tackles", y = "Average Fantasy Points")+
geom_point()
RB_broken_tackles## `geom_smooth()` using formula 'y ~ x'
- Now I will find each correlation for each of these stats in relation to fantasy points.
cor_avg_rush_att = cor.test(RB_totals$AVG_Rush_Att, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rush_yds = cor.test(RB_totals$AVG_Rush_Yds, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rush_td = cor.test(RB_totals$AVG_Rush_TD, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rush_long = cor.test(RB_totals$AVG_Rush_Long, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_targets = cor.test(RB_totals$AVG_Targets, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_rec = cor.test(RB_totals$AVG_Rec, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_rec_yds = cor.test(RB_totals$AVG_Receiving_Yards, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_rec_td = cor.test(RB_totals$AVG_Receiving_TD, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_yds_before_contact = cor.test(RB_totals$AVG_Yds_Before_Contact, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_yds_after_contact = cor.test(RB_totals$AVG_Yds_After_Contact, RB_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rb_broken_tackles = cor.test(RB_totals$AVG_Broken_Tackles, RB_totals$AVG_Total_DKP, method = "pearson")
RB_table = c(cor_avg_rush_att$estimate, cor_avg_rush_yds$estimate, cor_avg_rush_td$estimate, cor_avg_rush_long$estimate, cor_avg_rb_targets$estimate, cor_avg_rb_rec$estimate, cor_avg_rb_rec_yds$estimate, cor_avg_rb_rec_td$estimate, cor_avg_rb_yds_before_contact$estimate, cor_avg_rb_yds_after_contact$estimate, cor_avg_rb_broken_tackles$estimate)
row_names1 = c("Average Attempts", "Average Rushing Yards", "Average Rushing Touchdowns", "Average Longest Run", "Average Passing Targets", "Average Receptions", "Average Recieving Yards", "Average Recieving Touchdowns", "Average Yards Before Contact", "Average Yards After Contact", "Average Broken Tackles")
RB_table = as.data.frame(RB_table, row_names1)
RB_table## RB_table
## Average Attempts 0.6839729
## Average Rushing Yards 0.7976372
## Average Rushing Touchdowns 0.6288592
## Average Longest Run 0.6273653
## Average Passing Targets 0.7219403
## Average Receptions 0.7517249
## Average Recieving Yards 0.7629032
## Average Recieving Touchdowns 0.4680772
## Average Yards Before Contact 0.7184490
## Average Yards After Contact 0.7387496
## Average Broken Tackles 0.5616254- Next I will model the stats with the highest correlations.
# Between the fantasy values, rushing yards, receiving yards, and receptions have the biggest correlation to fantasy points.
model_rush_yds = lm(AVG_Total_DKP~AVG_Rush_Yds, RB_totals)
summary(model_rush_yds)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Rush_Yds, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -6.283 -2.279 -0.928 2.475 10.595
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.86256 0.90709 0.951 0.344
## AVG_Rush_Yds 0.21783 0.01708 12.754 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.439 on 93 degrees of freedom
## Multiple R-squared: 0.6362, Adjusted R-squared: 0.6323
## F-statistic: 162.7 on 1 and 93 DF, p-value: < 2.2e-16# y = 0.2178x + 0.8626 R-Squared = 0.6323 (Slightly Above Average)
model_receiving_yards_rb = lm(AVG_Total_DKP~AVG_Receiving_Yards, RB_totals)
summary(model_receiving_yards_rb)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Receiving_Yards, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.3084 -2.6325 -0.3575 2.1421 13.1407
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 6.0747 0.6099 9.96 2.49e-16 ***
## AVG_Receiving_Yards 0.3414 0.0300 11.38 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.686 on 93 degrees of freedom
## Multiple R-squared: 0.582, Adjusted R-squared: 0.5775
## F-statistic: 129.5 on 1 and 93 DF, p-value: < 2.2e-16# y = 0.3414 + 6.0747 R-Squared = 0.5775 (Average)
model_rec_rb = lm(AVG_Total_DKP~AVG_Rec, RB_totals)
summary(model_rec_rb)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Rec, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.727 -2.642 -0.274 2.210 13.778
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 5.3564 0.6806 7.87 6.33e-12 ***
## AVG_Rec 2.9568 0.2690 10.99 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.76 on 93 degrees of freedom
## Multiple R-squared: 0.5651, Adjusted R-squared: 0.5604
## F-statistic: 120.8 on 1 and 93 DF, p-value: < 2.2e-16# y = 2.9568x + 5.3564 R-Squared = 0.5604 (Average)
#Out of non-fantasy stats, yards after contact, before contact, and attempts
model_rush_before = lm(AVG_Total_DKP~AVG_Yds_Before_Contact, RB_totals)
summary(model_rush_before)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Yds_Before_Contact, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -9.1274 -2.8343 -0.5193 2.6859 10.8394
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.18656 1.02158 2.140 0.0349 *
## AVG_Yds_Before_Contact 0.36065 0.03621 9.961 2.48e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.966 on 93 degrees of freedom
## Multiple R-squared: 0.5162, Adjusted R-squared: 0.511
## F-statistic: 99.22 on 1 and 93 DF, p-value: 2.479e-16# y = 0.3607x + 2.1866 R-Squared = 0.511 (Average)
model_rush_after = lm(AVG_Total_DKP~AVG_Yds_After_Contact, RB_totals)
summary(model_rush_after)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Yds_After_Contact, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.6379 -2.5878 -0.3584 2.7851 9.9071
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.07624 0.89083 3.453 0.000836 ***
## AVG_Yds_After_Contact 0.36640 0.03466 10.570 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.843 on 93 degrees of freedom
## Multiple R-squared: 0.5458, Adjusted R-squared: 0.5409
## F-statistic: 111.7 on 1 and 93 DF, p-value: < 2.2e-16# y = 0.3664x + 3.0762 R-Squared = 0.5409 (Average)
model_rush_att = lm(AVG_Total_DKP~AVG_Rush_Att, RB_totals)
summary(model_rush_att)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Rush_Att, data = RB_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -9.4399 -2.5408 -0.5751 2.5571 11.2939
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -0.6459 1.4116 -0.458 0.648
## AVG_Rush_Att 1.0266 0.1135 9.042 2.18e-14 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.159 on 93 degrees of freedom
## Multiple R-squared: 0.4678, Adjusted R-squared: 0.4621
## F-statistic: 81.75 on 1 and 93 DF, p-value: 2.179e-14# y = 1.0266x - 0.6459 R-Squared = 0.4621 (Average)- Finally I will create the data frame for WR and TE stats.
WR_TE_df1 = sqldf("select pos, player, team, targets, rec, rec_yds, rec_td, rec_long, fumbles_lost, designed_rush_att, comb_pass_rush_play, comb_pass_play, Total_DKP
from df
where pos = 'WR' or pos = 'TE'")
WR_TE_totals = sqldf("select pos, player, team, avg(targets), avg(rec), avg(rec_yds), avg(rec_td), avg(rec_long), avg(fumbles_lost), avg(designed_rush_att), avg(comb_pass_rush_play), avg(comb_pass_play), avg(Total_DKP)
from WR_TE_df1
where targets > 2
group by player")- Now I will rename all the WR and TE columns so they can be used in SQL.
WR_TE_totals = rename(WR_TE_totals, 'AVG_Total_DKP' = 'avg(Total_DKP)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Targets' = 'avg(targets)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Rec' = 'avg(rec)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Rec_Yds' = 'avg(rec_yds)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Rec_TD' = 'avg(rec_td)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Rec_Long' = 'avg(rec_long)')
WR_TE_totals = rename(WR_TE_totals, 'AVG_Rec_Passing_Rushing_Plays' = 'avg(comb_pass_rush_play)')- Now I will graph each statistic in relation to fantasy points.
WR_TE_targets <- ggplot(WR_TE_totals, aes(x = AVG_Targets, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Targets and Fantasy Points", x = "Average Targets", y = "Average Fantasy Points")+
geom_point()
WR_TE_targets## `geom_smooth()` using formula 'y ~ x'
WR_TE_rec <- ggplot(WR_TE_totals, aes(x = AVG_Rec, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Receptions and Fantasy Points", x = "Average Rec", y = "Average Fantasy Points")+
geom_point()
WR_TE_rec## `geom_smooth()` using formula 'y ~ x'
WR_TE_rec_yds <- ggplot(WR_TE_totals, aes(x = AVG_Rec_Yds, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Recieving Yards and Fantasy Points", x = "Average Recieving Yards", y = "Average Fantasy Points")+
geom_point()
WR_TE_rec_yds## `geom_smooth()` using formula 'y ~ x'
WR_TE_rec_td <- ggplot(WR_TE_totals, aes(x = AVG_Rec_TD, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Recieving Touchdowns and Fantasy Points", x = "Average Recieving Touchdowns", y = "Average Fantasy Points")+
geom_point()
WR_TE_rec_td## `geom_smooth()` using formula 'y ~ x'
WR_TE_rec_long <- ggplot(WR_TE_totals, aes(x = AVG_Rec_Long, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Longest Catch and Fantasy Points", x = "Average Longest Catch", y = "Average Fantasy Points")+
geom_point()
WR_TE_rec_long## `geom_smooth()` using formula 'y ~ x'
WR_TE_passing_rushing_plays <- ggplot(WR_TE_totals, aes(x = AVG_Rec_Passing_Rushing_Plays, y = AVG_Total_DKP))+
geom_smooth(method = "lm", se = TRUE)+
labs(title = "Relationship Between Average Passing and Running Plays and Fantasy Points", x = "Average Passing and Running Plays", y = "Average Fantasy Points")+
geom_point()
WR_TE_passing_rushing_plays## `geom_smooth()` using formula 'y ~ x'
- I will now find the correlation for each statistic in relation to fantasy points and create a table.
cor_avg_targets = cor.test(WR_TE_totals$AVG_Targets, WR_TE_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rec = cor.test(WR_TE_totals$AVG_Rec, WR_TE_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rec_yds = cor.test(WR_TE_totals$AVG_Rec_Yds, WR_TE_totals$AVG_Total_DKP, method = "pearson")
cor_avg_td = cor.test(WR_TE_totals$AVG_Rec_TD, WR_TE_totals$AVG_Total_DKP, method = "pearson")
cor_avg_rec_long = cor.test(WR_TE_totals$AVG_Rec_Long, WR_TE_totals$AVG_Total_DKP, method = "pearson")
cor_avg_pass_rush_plays = cor.test(WR_TE_totals$AVG_Rec_Passing_Rushing_Plays, WR_TE_totals$AVG_Total_DKP, method = "pearson")
WR_TE_table = c(cor_avg_targets$estimate, cor_avg_rec$estimate, cor_avg_rec_yds$estimate, cor_avg_td$estimate, cor_avg_rec_long$estimate, cor_avg_pass_rush_plays$estimate)
row_name = c("Average Targets", "Average Receptions", "Average Recieving Yards", "Averager Recieving Touchdowns", "Average Longest Catch", "Average Run and Pass Plays")
WR_TE_table = as.data.frame(WR_TE_table, row_name)
WR_TE_table## WR_TE_table
## Average Targets 0.8073433
## Average Receptions 0.8496987
## Average Recieving Yards 0.9065176
## Averager Recieving Touchdowns 0.6500585
## Average Longest Catch 0.7132301
## Average Run and Pass Plays 0.1365477- Now we will model the statistics with the highest correlations.
# The fantasy statistic with the highest correlation is receiving yards.
model_for_rec_yds_wr_te = lm(AVG_Total_DKP~AVG_Rec_Yds, WR_TE_totals)
summary(model_for_rec_yds_wr_te)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Rec_Yds, data = WR_TE_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -5.6574 -1.2725 -0.2365 0.8408 12.4250
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.286414 0.263472 4.883 1.86e-06 ***
## AVG_Rec_Yds 0.199654 0.005857 34.087 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.906 on 252 degrees of freedom
## Multiple R-squared: 0.8218, Adjusted R-squared: 0.8211
## F-statistic: 1162 on 1 and 252 DF, p-value: < 2.2e-16# y = 0.1997x + 1.2864 R-Squared = 0.8211 (Good)
# Some non-calculated stat to consider is targets
model_for_target_wr_te = lm(AVG_Total_DKP~AVG_Targets, WR_TE_totals)
summary(model_for_target_wr_te)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Targets, data = WR_TE_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.0300 -1.7669 -0.5047 1.4713 7.8193
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.17380 0.50990 -2.302 0.0221 *
## AVG_Targets 1.95064 0.08981 21.719 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.664 on 252 degrees of freedom
## Multiple R-squared: 0.6518, Adjusted R-squared: 0.6504
## F-statistic: 471.7 on 1 and 252 DF, p-value: < 2.2e-16# y = 1.9506x -1.1738 R-Squared = 0.6518 (Above Average)
model_long_rec_wr_te = lm(AVG_Total_DKP~AVG_Rec_Long, WR_TE_totals)
summary(model_long_rec_wr_te)##
## Call:
## lm(formula = AVG_Total_DKP ~ AVG_Rec_Long, data = WR_TE_totals)
##
## Residuals:
## Min 1Q Median 3Q Max
## -9.821 -1.872 -0.399 1.818 14.251
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.71044 0.50945 3.357 0.000908 ***
## AVG_Rec_Long 0.39444 0.02442 16.153 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.165 on 252 degrees of freedom
## Multiple R-squared: 0.5087, Adjusted R-squared: 0.5067
## F-statistic: 260.9 on 1 and 252 DF, p-value: < 2.2e-16# y = 0.3944x + 1.7104 R-Squared = 0.5067 (Average)Part 2 conclusion:
When selecting a fantasy team:
At QB use:
y = 7.2347x + 6.9655 where x is the number of passing touchdowns.
This is the model that is most likely to give an accurate estimate.
To have an extra advantage on your opponents you can use y = 0.2676x -6.6782, where x is passer rating
At RB, the most reliable model is:
y = 0.2178x + 0.8626, where x is rushing yards
For an extra advantage you should look for physical runners who gain a lot of yards after contact
To predict their points you can use y = 0.3664x + 3.0762 where x is yards after contact
For WR and TE the best prediction model is:
y = y = 0.1997x + 1.2864 where x is receiving yards.
It is also important to look at how much their QB likes them using y = 2.7073x - 0.0067 where x is the number of targets.
Overall this project should help fantasy players look for which specific stats can help them select players who will put up good fantasy points in the future.